Insertion sites of the OSK lentivirus for each iPSC clone are indicated on a representation of the 20 mouse chromosomes. Sites of integration were determined by whole genome sequencing. Each iPSC line had 1–5 lentiviral insertion sites. The first number next to each insertion site refers to the experiment, and the second refers to the clone. Experiment 1 insertion sites are indicated in green, experiment 2 in red, and experiment 3 in blue.

Panel A. Total number of SNVs (including a small fraction of loss of heterozygosity sites) detected in each iPSC clone compared to its parental fibroblasts, related to Figure S2.
Panel B. Total number of SNVs detected in the coding regions of each iPSC clone compared to its parental fibroblasts, listed in Table S1. Pathway analysis of these SNVs is shown in Table S5.

Panel A. Variant allele frequency plots of all SNVs for each of the iPSC clones from experiments 1 and 2. Two plots are shown for each clone: kernel density (top) and iPSC variant allele frequencies by sequence coverage (bottom). The traditional kernel density estimation (KDE, Experimental Procedures) reveals that each iPSC sample is comprised of a single dominant clone, with no apparent subclones.
Panel B. Relationship of the total SNVs detected within each iPSC clone. No shared SNVs were detected in any clone in either experiment. Overlapping regions of the Venn diagram with no numbers contain no SNVs.

Panel A. Relationship of the SNVs detected within the coding regions of each iPSC clone. Private and shared mutations were found in all four clones. Overlapping regions of the Venn diagram with no numbers contain no SNVs. Pathway analysis of these SNVs is shown in Table S5.
Panel B. Relationship of the total SNVs (including a small fraction of loss of heterozygosity sites) for each iPSC clone. Overlapping regions of the Venn diagram with no numbers contain no SNVs. Shared SNVs are listed in Table S2 and shared indels are listed in Table S3.
Panel C. An identical structural variant was identified in all four iPSC clones from experiment 3, spanning ~130,000 bp on chromosome 6.
Panel D. Detection of shared variants in rare cells from the parental MEF pool. Regions containing the SNVs within the Apaf1 and Sbno2 genes were amplified from the Gusb −/− MEF starting pool from experiment 3, and these amplicons were directly ligated into the pCR2.1 vector. The ligation products were transformed into E. coli; ampicillin resistant colonies were counted on each 10 cm plate, and pooled. Plasmid DNA was prepared from each pool, and digested with StuI and EcoRI to detect the novel StuI sites generated by the Apaf1 G16A and Sbno2 A3783G variants. Digestion products were separated on 5% polyacrylamide gels, transferred to nitrocellulose, and analyzed by Southern blotting. Expected fragment lengths are indicated below the Southern blots. * indicates the variant-specific digestion products. Quantification data from Illumina sequencing validation is shown in Table S4.

Left and Center panels: Variant allele frequencies for the shared and private SNVs from the iPSC clones of experiment 3 were plotted. Two plots are shown for each clone: kernel density (top) and iPSC variant allele frequencies by sequence coverage (bottom). The shared SNVs have a single dominant clone with a peak variant allele frequency of ~50% (left panels). The private SNVs in each clone also have a major peak of variant allele frequency of ~50%, but the suggestion of second peak at ~25% is suggested in clones 1, 2 and 6. The second peak was confirmed only for clone 1 (red arrowhead).
Right panels: Validation of variant allele frequencies in iPSC clones. Using whole genome sequencing data, we selected ten variants with predicted VAFs of 50% vs. 25% for each of the iPSC clones, amplified all variants by PCR, and then performed deep sequencing on the 454-FLX platform. Variant allele frequencies (VAFs) were plotted for each clone for whole genome sequencing (X axis, VAF by WGS) vs deep sequencing (Y axis, VAF by deep 454). For clone 1, the variants with VAFs of ~25% were statistically confirmed by deep sequencing (p=0.000045). For the other three clones, the variants with predicted 25% VAF were not confirmed (p>0.05 for all three).

Panel A. Model for the genetic variants detected in experiments 1 and 2. Each cell has a unique set of pre-existing background mutations, which are “captured” by the expansion and cloning of single cells with reprogramming. Not all cells transduced with the reprogramming vector yield iPSC clones, as suggested by the X’s. It is possible that the background mutations in some cells could contribute to reprogramming fitness.
Panel B. Model for the data obtained for experiment 3. Preexisting background mutations in rare cells mark their fitness for reprogramming. Private mutations must be acquired after the shared mutations, and could have arisen at different times: 1) between the time when the shared mutations were acquired and when reprogramming occurred, or 2) at the time of reprogramming, or 3) after reprogramming. Some mutations arising after reprogramming may provide a selective advantage for the outgrowth of subclones.